ReviewAid

ReviewAid is a free, open-source tool for AI-assisted full-text screening and data extraction in systematic reviews. It suits researchers who want to tailor screening criteria and extraction fields, and are prepared to review the tool’s work. Its appeal is configurable assistance without a subscription; it is not a replacement for human judgment.

Overview

ReviewAid applies user-defined PICO inclusion and exclusion criteria to full-text papers, then extracts fields chosen by the user. Confidence scores and low-confidence flags can help reviewers prioritize checks, but the maker positions the software as supplementary: screening and extraction decisions still require human oversight. Reference deduplication is supported, while blinded screening is not.

The Apache 2.0 license allows researchers to run the software locally or use its browser-based Streamlit app without installation. Local setup requires Python 3.12 or later and Streamlit. The hosted app may take about 30 seconds to initialize. For privacy-sensitive work, Ollama runs inference on the local machine without an external API key, keeping paper text there. Selecting a cloud provider sends paper text to that provider for the run. The maker says API keys are not stored.

Key features

Configurable screening and extraction

Researchers can define the PICO criteria for full-text screening and the data fields to extract. This flexibility can suit reviews with focused or project-specific questions, but puts the responsibility for well-formed criteria and fields on the review team. Results still need checking rather than automatic acceptance.

Confidence, exports, and batch sizes

Results can be downloaded as CSV, XLSX, or DOCX, giving teams several formats for further review and documentation. Confidence scores and low-confidence flags provide a useful way to direct manual attention, though they do not guarantee that high-confidence results are correct. The repository instructions cap extraction submissions at 20 articles and full-text screening submissions at 10, so larger reviews need to be handled in batches. The site describes batch processing of up to 20 papers, but screening has the lower stated limit.

AI providers and local inference

ReviewAid supports OpenAI, Anthropic, DeepSeek, Cohere, Z.ai, and local Ollama. That choice lets teams select a provider or run inference locally. Ollama is the clearest fit when keeping paper text off external services is a priority; cloud-provider runs transmit that text to the selected provider.

Pricing

ReviewAid is free: its Free plan costs nothing and is open source under Apache 2.0. There is no paid tier to weigh against the free version. The practical constraints are the per-submission batch limits, the need for human review, and—if using a cloud provider—the transfer of paper text for processing. The app can be run locally, but local setup requires Python 3.12 or later and Streamlit.

Platforms

ReviewAid supports Linux, macOS, Windows, web, and self-hosted use. The web app avoids installation, while local deployment gives researchers a way to run the workflow on their own machine. Ollama inference can keep paper text local; using a cloud provider instead sends it to that provider for the run.

Who it's for

ReviewAid is aimed at researchers, especially those working on evidence synthesis and systematic reviews. It is a good fit for teams that need configurable full-text PICO screening and extraction, can verify AI-generated results, and are comfortable working within the batch limits. Researchers who need blinded screening, want to hand decisions over to automation, or cannot use cloud processing and do not want to manage a local setup should look elsewhere.

Pros and cons

Pros

• Free and open source under Apache 2.0, with no subscription cost.
• Custom PICO screening and extraction fields accommodate review-specific questions.
• Local Ollama inference keeps paper text on the machine and requires no external API key.
• Confidence indicators and CSV, XLSX, and DOCX exports support checking and downstream work.

Cons

• Human reviewers must still verify screening and extraction; the tool is supplementary.
• Full-text screening is limited to 10 articles per submission, and extraction to 20, adding batching work for larger reviews.
• Cloud-provider processing sends paper text to the selected provider.
• Blinded screening is not supported, and the hosted app may have cold starts of about 30 seconds.

Alternatives

Universal SLR Assistant is another free, open-source option for Linux, macOS, web, and Windows; its listed free plan notes that cloud AI provider API fees may apply. Choose it if that option better fits your workflow or platform needs.

ASReview is free and open source, with Linux, macOS, web, Windows, API, and self-hosted support. It is an alternative for readers who want a tool with those deployment options and no licensing fees or subscriptions, while running their own hardware or server may incur costs.

RevMan is freemium, offers a free plan and free trial, and supports Linux, macOS, web, and Windows. Consider it if those terms or platform options suit your needs better.

Rayyan is freemium and supports Android, iOS, web, and API access. Its free plan is limited to three active reviews, two free reviewers, one sample, and limited mobile app access; choose it if those limits and mobile support fit your work.

Bango is a free option for Linux, macOS, and Windows, described as available to researchers without a subscription, paywall, or limits. It may suit readers seeking that stated unrestricted model.

Parsifal is a free web option whose AI assistant is optional and uses the reviewer's own Anthropic or OpenAI API key.

Silvi is a web-based freemium alternative with a free Basic plan and a paid Rapid plan at €19 a month. Consider it if its plan structure suits your needs.

Covexe is a freemium web alternative with a free plan.

For a broader comparison, see Systematic Review Software.

Verdict

Choose ReviewAid if you want free, configurable AI assistance for full-text screening and extraction, and can keep human reviewers responsible for checking the output. Its local Ollama option is a strong reason to consider it when paper privacy matters. Look elsewhere if you need blinded screening, larger screening batches without extra handling, or a workflow that does not depend on manual verification.